tenstorrent / tt-buda

Tenstorrent TT-BUDA Repository

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

tt logo

TT-Buda is a software stack that can compile AI/ML models from several different frameworks, and execute them on Tenstorrent hardware.


Introduction

The TT-Buda software stack can compile AI/ML models from several different frameworks such as PyTorch and Tensorflow, and execute them in many different ways on Tenstorrent hardware.

Note on terminology:

TT-Buda is the official Tenstorrent AI/ML compiler stack and PyBuda is the Python interface for TT-Buda. PyBuda allows users to access and utilize TT-Buda's features directly from Python. This includes directly importing model architectures and weights from PyTorch, TensorFlow, ONNX, and TFLite.

Model Demos

Model demos are now part of a separate repo:

https://github.com/tenstorrent/tt-buda-demos

Docs

See: Docs

Build

https://docs.tenstorrent.com/tenstorrent/v/tt-buda/installation

Env setup

Set LD_LIBRARY_PATH to the location of third_party/budabackend/build/lib - preferrably the absolute path to allow scripts to find them from anywhere.

Silicon

See README.silicon.md for details on how to run on silicon.

About

Tenstorrent TT-BUDA Repository

License:Other


Languages

Language:Python 63.4%Language:C++ 35.9%Language:Makefile 0.3%Language:Shell 0.3%Language:Jinja 0.1%